The therapeutic effect of balloon dilatation with different duration for biliary duct calculi: A network meta-analysis.

Objective: To systematically evaluate the application effect of endoscopic papillary balloon dilatation (EPBD) with different balloon dilatation duration for biliary duct calculi, and find the most appropriate dilatation duration for EPBD using a network meta-analysis. Materials and
Methods: PubMed, Embase and Cochrane Library databases were searched for relevant randomised controlled trials (RCTs) published up to August 2020. Node split, consistency and inconsistency models analysis were all conducted in network meta-analysis.
Results: Eighteen RCTs with 2256 participants were finally analysed. EPBD was divided into four categories based on balloon dilatation duration, including EPBD (P0.5), EPBD (>0.5, ≤1), EPBD (1, ≤2) and EPBD (>2, ≤5). Compared with EPBD (>0.5, ≤1), EPBD (>2, ≤5) had a lower risk of early complications (odds ratio [OR] = 0.23, 95% credible interval [CI] = 0.05-0.96) and post-endoscopic procedure pancreatitis (PEP) (OR = 0.17, 95% CI = 0.03-0.72). Endoscopic sphincterotomy (EST) tended to have less need for mechanical lithotripsy (OR = 0.37, 95% CI = 0.16-0.88) and PEP (OR = 0.26, 95% CI = 0.08-0.71) than EPBD (>0.5, ≤1). EPBD (>2, ≤5) was the safest endoscopic procedure with respect to early complications (surface area under cumulative ranking curves [SUCRA] = 79.0) and PEP (SUCRA = 85.3). In addition, EPBD (>2, ≤5) and EST had the highest probability of being the best (SUCRA = 82.6) and the worst (SUCRA = 10.8), respectively, regarding late complications.
Conclusion: EPBD and EST are two methods used to treat uncomplicated choledocholithiasis (stone diameter <10 mm and stone number <3). The extension of balloon dilatation duration has no significant influence on successful stone removal in the first endoscopic session or preventing the need for mechanical lithotripsy. However, it can reduce the risk of early complications, especially PEP. What's more, EPBD seems to have less late complications compared with EST, and the effect of prolonged balloon dilatation duration on late complications still needs to be further explored. Therefore, 2-5 min is the recommended dilatation duration range for EPBD using balloon with ≤10 mm diameter. Further research based on a specific population and with a longer follow-up time are needed.

INTRODUCTION

Endoscopic treatment has become the first choice for biliary duct calculi,[123] due to its advantages of less invasive than surgery, high repeatability. Endoscopic sphincterotomy (EST) and endoscopic papillary balloon dilatation (EPBD) are the two main methods of endoscopic lithotomy.[456]

EPBD, first reported in 1983, has been relatively mature after nearly four decades of development. In addition, endoscopic papillary large balloon dilatation (EPLBD) and EST with balloon dilatation (ESBD) were developed from EPBD, which have been used to treat large or complex stones.[789]

The action mechanism of EPBD is temporarily loosen the sphincter by expanding the balloon, to create space for stone extraction. Moreover, the balloon with different sizes can be selected according to the diameters of stone and common bile duct (CBD).[710] A recent cumulative meta-analysis has shown that EPLBD causes a similar effect and less tissue damage compared with EST. However, the efficacy of EPBD that using a smaller diameter balloon (≤10 mm) is still controversial. Although the use of EPBD can retain the function of the Oddi sphincter, as several studies reported, it increases the difficulty and failure rate of stone extraction. In particular, post endoscopic procedure pancreatitis (PEP), which has been the focus of attention,[1111213] limited the clinical application of EPBD.

Liao et al.[1415] found that the duration of balloon dilatation would affect the efficacy of EPBD, and an extended duration of balloon dilatation could reduce the incidence of PEP. Over the past decade, few studies compared the efficacy of two different duration EPBD, let alone randomised controlled trials (RCTs). Moreover, the duration of balloon dilatation reported varied widely-from 15 s to 5 min.[16171819] Therefore, we collected published RCTs that comparing two different duration EPBD, or comparing EPBD with EST. Then, we analysed the efficacy and complications of EPBD with different balloon dilatation duration using network meta-analysis, to find the most appropriate dilatation duration and guide the clinical treatment.

MATERIALS AND METHODS

This study was designed and implemented according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement,[20] and it has been registered in the PROSPERO database (Registration no. CRD42020203928).

Search strategy

The PubMed, EMBASE and Cochrane Library databases were searched for studies of EPBD for CBD stones until August 2020, and language was limited to English. Subject words and random words were combined for retrieval, which included 'sphincterotomy, endoscopic’, ‘EST’, ‘EPBD’, ‘gallstones’ and ‘CBD stone’. Besides, references in the literature were also manually searched. The preliminary screening of collected studies was conducted by reading titles and abstracts. Then, the full text was read to identify the studies that meet the inclusion criteria.

Inclusion and exclusion criteria

The studies meeting the following criteria were included: (1) RCTs; (2) studies including patients with one or more CBD stones; (3) experimental group and control group were treated with EPBD and EST, or two different duration EPBD; (4) studies published in English; (5) basic data of patients and endoscopic procedures were both provided in studies. The exclusion criteria were as follows: (1) cohort studies, case-control studies, review articles or case reports; (2) repeated studies; (3) patients treated with EPBD combined with EST; (4) patients with recurrent choledocholithiasis; (5) implanted balloon with large diameter(>10 mm); (6) duration of balloon dilatation were not provided.

Data extraction and quality assessment

Two investigators independently carried out data extraction and quality assessment. Further discussions were held, or a third investigator consulted when both sides held different opinions. The following data were extracted from initial studies: Study characteristics (first author, year, country and follow-up time), patient characteristics (sample, gender, median age), stone characteristics (number, median diameter), endoscopic technique characteristics (type, balloon diameter, dilatation duration) and treatment outcomes (successful stone removal in the first endoscopic session, need for mechanical lithotripsy, early complications, PEP, late complications, stone recurrence). The quality assessment of the selected trials was implemented by using the Cochrane risk of bias tool.[21]

Statistical analysis

Bayesian network meta-analysis, which is based on the Markov chain Monte Carlo method, was implemented to incorporate direct information and indirect information. Since the type of outcome data were all binary, odds ratio (OR) with 95% credible interval (CI) were calculated. Subsequently, the Bayesian random-effects regression model was created in GeMTC 0.14.3 software: Four Markov chains were initially set up for modelling, and the model was updated with 100,000 simulation iterations and a burn-in of 50,000 iterations. A node-splitting analysis was utilised to calculate the inconsistency, demonstrating statistical inconsistencies with P < 0.05 and 95% CI beyond the null value.[22] We utilised both consistency and inconsistency model approaches to test the quality of the results. The Stata SE 15 software was used to draw network diagrams and compute the cumulative ranking probabilities using the surface area under cumulative ranking curves (SUCRA). The possibility of publication bias presented in a funnel plot and literature quality assessment were implemented by using Review Manager 5.3 software (The Cochrane Collaboration, Software Update, Oxford, UK).

RESULTS

Study selection and study characteristics

A total of 472 studies were identified based on retrieval strategy, and 23 repetitive articles were excluded by using NoteExpress software. After reading the titles and abstracts, 109 studies were left. Out of the remaining 109 studies, 18[111214161723242526272829303132333435] were included in this network meta-analysis. The flow diagram of literature screening is shown in Figure 1.

Figure 1

Flow diagram of literature screening and selection

There were 2256 patients involved in the 20 RCTs. 2[1625] of the 20 studies reported outcome indicators for the same sample at different times. All included studies were dual-arm trials, among which 14 trials[1112232425262728293031323334] compared EPBD and EST, and 4 trials[14161735] compared two EPBD groups with different balloon dilatation duration. In terms of country, there were nine studies from Japan,[112325282930313334] taking up almost half of the included studies. All studies were divided into six categories according to endoscopic methods and balloon dilatation duration, including EPBD (P0.5) versus EST,[232425] EPBD (>0.5, ≤1) versus EST,[2627] EPBD (>1, ≤2) versus EST,[11122829303132] EPBD (>2, ≤5) versus EST,[3334] EPBD (P0.5) versus EPBD (>0.5, ≤1),[1735] and EPBD (>0.5, ≤1) versus EPBD (>2, ≤5).[1416] The characteristics of included studies were shown in Table 1. Network graphs for outcomes were plotted to show the comparative relationship between each intervention, with nodes representing the endoscopic approaches [Figure 2].

Table 1

Characteristics of included studies

Author (year)	Country	Treatment	Number of patient	Gender (male/female)	Age (years)	Number of stone	Diameter of stone (mm)	Diameter of ballon (mm)	Follow-up
EPBD (≤0.5) versus EST
Fujita (2003)	Japan	EPBD (15 s)	138	75/63	66.8 (26-93)	2.4±2.5	7.0±3.1	8	-
		EST	144	92/52	68.4 (31-93)	2.4±2.9	7.3±3.4		-
Vlavianos (2003)	UK	EPBD (30 s)	103	25/78	60.8±20.5	-	-	10	12 months
		EST	99	35/64	61.9±18.3	-	-		12 months
Yasuda (2010)	Japan	EPBD (15 s)	138	75/63	68.5 (26-93)	1 (1-16)	6.5 (2-15)	4-6-8	6.7 years
		EST	144	92/52	71 (31-93)	1 (1-24)	7 (2-16)		6.7 years
EPBD (>0.5-≤1) versus EST
Bergman (1997)	Netherlands	EPBD (45-60 s)	101	43/58	72 (27-98)	2 (1-14)	10 (3-36)	8	6 months
		EST	101	45/56	71 (29-96)	1 (1-15)	9 (4-27)		6 months
Disario (2004)	USA	EPBD (60 s)	117	41/76	47±19	1 (1-100)	6 (0.5-10)	8	48±88 days
		EST	120	31/89	54±19	1 (1-10)	5 (0.5-14)		47±42 days
EPBD (>1-≤2) versus EST
Arnold (2001)	Germany	EPBD (60 s×2)	30	11/19	54.2±18.5	1.6±1.1	7.0±3.5	8
		EST	30	13/17	58.5±18.5	1.8±1.5	10±4.7
Yasuda (2001)	Japan	EPBD (60 s×2)	35	16/19	69.5 (42-86)	3.7 (1-16)	12.4 (4-24)	8	12 months
		EST	35	21/14	69.4 (43-88)	3.3 (1-16)	12.3 (5-24)		12 months
Natsui (2002)	Japan	EPBD (2 min)	70	33/37	64.5 (23-87)	2.7 (1-15)	9.2 (3-22)	8	29 (12-54) months
		EST	70	33/37	67.1 (38-88)	2.6 (1-15)	9.7 (3-17)		30 (12-54) months
Takezawa (2004)	Japan	EPBD (2 min)	46	32/14	70 (40-90)	1 (1-7)	10 (1-35)	8	12 months
		EST	45	30/15	69 (41-93)	1 (1-7)	11 (3-27)		12 months
Tanaka (2004)	Japan	EPBD (2 min)	16	10/6	67.2 (50-78)	2 (1-12)	10.2±3.5	8	61.5 (54-76) months
		EST	16	13/3	70.6 (49-87)	2 (1-4)	12.4±6.0		62.0 (2-74) months
Watanabe (2007)	Japan	EPBD (2 min)	90	51/39	69.1±13.1	2.7±2.8	8.1±3.2	8	-
		EST	90	49/41	70.2±8.1	2.5±2.7	7.7±2.9		-
Seo (2014)	Korea	EPBD (90-120 s)	62	27/35	32.1±7.3	1.5 (1-5)	7.2±2.1	6-10	35.4 months
		EST	70	32/38	33.2±5.8	1.8 (1-8)	7.6±3.1		35.4 months
EPBD (>2-≤5) versus EST
Minami (1995)	Japan	EPBD (3 min)	20	13/7	64±11.2	-	<12<12	8	21.5±6.2 months
		EST	20	9/11	71.3±14	-			23.1±5.0 months
Ochi (1999)	Japan	EPBD (60 s ×3)	55	34/21	62.6±15.9	2.1±1.9	8.1±3.4	8	23 (4-42) months
		EST	55	31/24	66.3±14.3	1.7±1.2	8.8±4.2		23 (4-42) months
EPBD (≤0.5) versus EPBD (>0.5-≤1)
Bang (2010)	Korea	EPBD (20 s)	35	16/19	63.3±13.6	-	8.2±3.3	9.6±2.4	-
		EPBD (60 s)	35	19/16	66.2±17.4	-	8.1±3.5	9.7±2.6	-
Bang (2015)	Korea	EPBD (20 s)	109	58/51	62.0±16.9	-	6.5±2.7	6-10	-
		EPBD (60 s)	119	74/45	63.7±16.6	-	6.9±2.9		-
EPBD (>0.5-≤1) versus EPBD (>2-≤5)
Liao (2010)	China	EPBD (1 min)	86	41/45	64.7±15.9	-	6.0 (2-23)	10	1 months
		EPBD (5 min)	84	44/42	61.2±17.4	-	6.3 (2-30)		1 months
Kuo (2017)	China	EPBD (1 min)	86	41/45	64.7±15.9	-	6.0 (2-23)	10	7.4 (6.6-8) months
		EPBD (5 min)	84	44/42	61.2±17.4	-	6.3 (2-30)		6.9 (6.7-7.7) months

Data are reported as numbers, mean±SD, median (range). Treatment are shown as EST or EPBD (balloon dilatation duration). EPBD: Endoscopic papillary balloon dilatation, (≤0.5)/(>0.5-≤1)/(>1-≤2)/(>2-≤5), balloon dilatation duration (min), EST: Endoscopic sphincterotomy, SD: Standard deviation

Figure 2

(a) Network graphs of included studies for successful stone removal (1st session), (b) need for mechanical lithotripsy, (c) early complications, (d) post-endoscopic procedure pancreatitis, (e) late complications, (f) stone recurrence

Quality assessment

The risk of bias in the included studies was mostly low, and there was no high risk. 8 studies did not explain the randomisation methods,[1112232931333435] and 5 studies did not explain whether allocation concealment was used or what the specific concealment method was.[1231333435] The risk of reporting bias in five studies was unclear for the following reasons: four studies did not provide data about bleeding or perforation,[12243334] and one study had an obscure definition of bleeding.[27] The bias assessment results are shown in Figure 3.

Figure 3

Results of bias assessed with the Cochrane risk-of-bias tool

Network meta-analysis

Successful stone removal in the first endoscopic session

Regarding successful stone removal in the first endoscopic session, there was no statistically significant difference in all comparisons [Table 2]. EPBD (>1, ≤2), EST and EPBD (>2, ≤5) had high probabilities of being ranked first (SUCRA = 75.7), second (SUCRA = 74.4) and last (SUCRA = 8.1) respectively [Figure 4a]. The results showed that the extension of balloon dilatation time had no significant effect on the success rate of single lithotomy.

Table 2

Efficacy estimates table from network meta-analysis: Mean odds ratio (95% credible interval)

EPBD (≤0.5)	0.46 (0.13-1.82)	0.95 (0.17-3.89)	0.23 (0.04-1.59)	1.12 (0.35-3.84)	a
	EPBD (>0.5-≤1)	2.08 (0.30-9.17)	0.51 (0.08-2.89)	2.44 (0.60-9.18)
		EPBD (>1-≤2)	0.24 (0.04-2.15)	1.17 (0.51-3.79)
			EPBD (>2-≤5)	4.82 (0.85-27.82)
				EST
EPBD (≤0.5)	2.35 (0.82-6.31)	1.47 (0.54-3.65)	1.68 (0.40-6.57)	0.88 (0.38-1.87)	b
	EPBD (>0.5-≤1)	0.62 (0.23-1.71)	0.72 (0.22-2.33)	0.37 (0.16-0.88)
		EPBD (>1-≤2)	1.14 (0.30-4.34)	0.59 (0.35-1.02)
			EPBD (>2-≤5)	0.52 (0.15-1.80)
				EST
EPBD (≤0.5)	1.34 (0.46-4.25)	0.82 (0.20-3.03)	0.31 (0.06-1.63)	0.67 (0.22-1.92)	c
	EPBD (>0.5-≤1)	0.61 (0.15-2.07)	0.23 (0.05-0.96)	0.50 (0.17-1.25)
		EPBD (>1-≤2)	0.38 (0.07-2.10)	0.81 (0.36-1.93)
			EPBD (>2-≤5)	2.15 (0.51-9.35)
				EST
EPBD (≤0.5)	1.09 (0.35-3.60)	0.46 (0.10-1.88)	0.19 (0.03-1.05)	0.29 (0.08-0.90)	d
	EPBD (>0.5-≤1)	0.42 (0.10-1.55)	0.17 (0.03-0.72)	0.26 (0.08-0.71)
		EPBD (>1-≤2)	0.40 (0.07-2.38)	0.62 (0.26-1.48)
			EPBD (>2-≤5)	1.53 (0.33-7.21)
				EST
EPBD (≤0.5)	1.30 (0.20-7.19)	1.16 (0.29-5.37)	0.65 (0.09-3.88)	2.09 (0.66-6.51)	e
	EPBD (>0.5-≤1)	0.92 (0.21-5.19)	0.51 (0.12-1.94)	1.62 (0.44-6.77)
		EPBD (>1-≤2)	0.56 (0.08-2.82)	1.79 (0.68-4.06)
			EPBD (>2-≤5)	3.20 (0.77-16.13)
				EST
EPBD (≤0.5)	2.75 (0.45-16.77)	2.12 (0.46-10.18)	1.62 (0.23-11.13)	2.42 (0.69-8.70)	f
	EPBD (>0.5-≤1)	0.76 (0.17-3.68)	0.59 (0.16-2.04)	0.90 (0.25-3.29)
		EPBD (>1-≤2)	0.75 (0.14-4.48)	1.16 (0.47-2.90)
			EPBD (>2-≤5)	1.54 (0.35-7.05)
				EST

EPBD: Endoscopic papillary balloon dilatation, EST: Endoscopic sphincterotomy

Figure 4

(a) Rank probabilities of the different endoscopic procedures for successful stone removal. (1st session), (b) need for mechanical lithotripsy, (c) early complications, (d) post endoscopic procedure pancreatitis, (e) late complications, (f) stone recurrence. A: EST, B: EPBD EPBD (≤0.5), C: EPBD (>0.5, ≤1), D: EPBD (>1, ≤2), E: EPBD (>2, ≤5). EPBD: Endoscopic papillary balloon dilatation, EST: Endoscopic sphincterotomy

Need for mechanical lithotripsy

As showed by the network meta-analysis, the need for mechanical lithotripsy was less common in EST compared with EPBD (>0.5, ≤1) (OR = 0.37, 95% CI = 0.16–0.88). Furthermore, EST tended to have less need for mechanical lithotripsy than EPBD (>1, ≤2) (OR = 0.59, 95% CI = 0.35–1.02), but the difference between treatments was not statistically significant [Table 2]. EPBD (≤0.5), EST and EPBD (>2, ≤5) had the high probabilities of being ranked first (SUCRA = 85.1), second (SUCRA = 84.1), and last (SUCRA = 16.7) respectively [Figure 4b]. Hence, the balloon dilatation duration exerts no significant effect on the need for mechanical lithotripsy.

Early complications and post endoscopic procedure pancreatitis

Compared with EPBD (>0.5, ≤1) (OR = 0.23, 95% CI = 0.05–0.96), EPBD (>2, ≤5) had a lower risk of early complications, which includes PEP, haemorrhage, perforation, jaundice, bile leakage and mortality [Table 2]. EPBD (>2, ≤5) and EST had the high probabilities of being ranked first (SUCRA = 79.0) and second (SUCRA = 64.0) respectively. EPBD (≤0.5) and EPBD (>0.5, ≤1) both had high probabilities of being ranked last [Figure 4c].

Compared with EPBD (>0.5, ≤1), EPBD (>2, ≤5) (OR=0.17, 95%CI=0.03–0.72) and EST (OR=0.26, 95%CI=0.08–0.71) both had a lower risk of PEP. Moreover, EST had less PEP (OR = 0.29, 95% CI = 0.08–0.90) than EPBD (≤0.5) [Table 2]. In rank probability analysis, EPBD (>2, ≤5), EST, EPBD (>1, ≤2), EPBD (>0.5, ≤1) and EPBD (≤0.5) had the high probabilities of being ranked first (SUCRA = 85.3), second (SUCRA = 77.8), third (SUCRA = 43.6), forth (SUCRA = 35.4) and last (SUCRA = 7.9), respectively [Figure 4d]. Based on the results of this analysis, we can find that prolonged balloon dilatation duration reduced the risk of early complications, especially PEP.

Late complications and stone recurrence

As results showed, there was no statistically significant difference in all comparisons regarding late complications, including stone recurrence, cholangitis, cholecystitis and liver abscess [Table 2]. The rank probability analysis showed that EPBD (>2, ≤5) had the highest probability of being best (SUCRA = 82.6), while EST had the highest probability of being worst (SUCRA = 10.8) [Figure 4e].

There was no significant difference amongst the various comparisons as far as the stone recurrence (the most common complication) was concerned. EPBD (>2, ≤5) (SUCRA = 59.8) and EPBD (≤0.5) (SUCRA = 87.3) both had the high probabilities of being ranked first, while EPBD (>2, ≤5) had the highest probability of being fourth (SUCRA = 33.3) [Figure 4f]. The balloon dilatation duration has no marked effect on late complications or stone recurrence.

Consistency and inconsistency model analysis

Consistency and inconsistency model analysis were both conducted based on each clinical parameter.

For the consistency model analysis, all PSRF values were 1.00, manifesting that all analyses had achieved good convergence. The results depicting inconsistency factors and variance calculations also indicated a high consistency indirectly [Table 3].

Table 3

Results of inconsistency factors and variance calculations for inconsistency analysis

Parameter	Inconsistency factors		Variance calculation
	Cycle	Median (95% CI)	SD, median (95% CI)
			Random effects	Inconsistency
Successful stone removal (1st session)	A, B, D, E	0.01 (−1.80-2.06)	0.77 (0.14-2.10)	0.89 (0.02-2.79)
	B, D, E	0.00 (−1.93-1.92)
Need for mechanical lithotripsy	A, B, D, E	−0.21 (−1.82-0.59)	0.34 (0.03-0.94)	0.57 (0.03-1.16)
	B, D, E	−0.00 (−1.13-1.02)
Early complications	A, B, D, E	0.01 (−1.41-1.46)	0.71 (0.27-1.41)	0.64 (0.03-1.67)
	B, D, E	−0.04 (−1.50-1.16)
PEP	A, B, D, E	0.21 (−1.15-2.60)	0.60 (0.05-1.56)	0.88 (0.04-2.54)
	B, D, E	0.03 (−1.42-1.87)
Late complications	B, D, E	0.09 (−1.18-1.74)	0.55 (0.07-1.40)	0.72 (0.04-1.52)
Stone recurrence	B, D, E	0.00 (−1.06-1.12)	0.36 (0.02-0.96)	0.48 (0.02-0.99)

In the inconsistency factors, when the median is close to 0, the difference between the direct consequence and the indirect consequence of the network meta-analysis is small, further indicating that the network meta-analysis is basically based on the consistency hypothesis. In the variance calculation, when the random effects standard deviation median is close to the inconsistency standard deviation median, the inconsistency probability of the network meta-analysis is scarce, further indicating that the consistency of the Network Meta-analysis is high. A, EPBD (≤0.5); B, EPBD (>0.5-≤1); D, EPBD (>2-≤5); E, EST; CI: Credible interval, PEP: Post-endoscopic procedure pancreatitis, EST: Endoscopic sphincterotomy, SD: Standard deviation, EPBD: Endoscopic papillary balloon dilatation

Node split

The node split models were constructed to assess inconsistencies between direct and indirect estimates in all comparisons. In this analysis, a large P value (P > 0.05) indicated that no significant inconsistency had been found. All P values in successful stone removal during the first endoscopic session, need for mechanical lithotripsy, early complications, PEP, late complications and stone recurrence groups presented a P > 0.05, demonstrating no potential inconsistent risk existed. Node split models are shown in Table 4.

Table 4

Results of node splitting models

Parameters	Name	Median (95% CI)			P
		Direct effect	Indirect effect	Overall
a	EPBD (≤0.5), EPBD (>0.5-≤1)	−0.66 (−2.52-1.23)	−0.86 (−3.45-1.72)	−0.78 (−2.06-0.60)	0.88
	EPBD (≤0.5), EST	0.07 (−1.59-1.76)	0.31 (−2.45-3.19)	0.11 (−1.06-1.35)	0.85
	EPBD (>0.5-≤1), EPBD (>2-≤5)	−0.59 (−3.38-2.03)	−0.78 (−3.86-2.30)	−0.68 (−2.48-1.06)	0.91
	EPBD (>0.5-≤1), EST	0.94 (−1.43-3.32)	0.85 (−1.23-2.83)	0.89 (−0.51-2.22)	0.95
	EPBD (>2-≤5), EST	1.66 (−0.86-4.19)	1.49 (−1.68-4.70)	1.57 (−0.16-3.33)	0.91
b	EPBD (≤0.5), EPBD (>0.5-≤1)	−0.10 (−1.81-1.64)	1.34 (0.11-2.68)	0.86 (−0.20-1.84)	0.2
	EPBD (≤0.5), EST	0.07 (−0.78-0.97)	−1.40 (−3.31-0.57)	−0.13 (−0.96-0.63)	0.17
	EPBD (>0.5-≤1), EPBD (>2-≤5)	−0.89 (−2.54-0.65)	0.67 (−1.27-3.15)	−0.33 (−1.53-0.84)	0.21
	EPBD (>0.5-≤1), EST	−1.04 (−2.19-0.06)	−0.84 (−2.64-0.68)	−0.99 (−1.82-−0.12)	0.82
	EPBD (>2-≤5), EST	−1.44 (−3.54-0.34)	0.13 (−1.57-2.07)	−0.65 (−1.88-0.59)	0.2
c	EPBD (≤0.5), EPBD (>0.5-≤1)	0.41 (−1.07-2.00)	0.18 (−1.69-2.08)	0.29 (−0.78-1.45)	0.83
	EPBD (≤0.5), EST	−0.48 (−1.92-0.90)	−0.23 (−2.18-1.69)	−0.40 (−1.50-0.65)	0.82
	EPBD (>0.5-≤1), EPBD (>2-≤5)	−1.61 (−3.70-0.41)	−1.28 (−3.67-0.85)	−1.47 (−2.99-−0.04)	0.82
	EPBD (>0.5-≤1), EST	−0.58 (−1.98-0.68)	−0.93 (−2.63-0.72)	−0.69 (−1.74-0.23)	0.73
	EPBD (>2-≤5), EST	0.64 (−1.33-2.73)	0.94 (−1.44-3.25)	0.76 (−0.68-2.24)	0.84
d	EPBD (≤0.5), EPBD (>0.5-≤1)	0.37 (−1.05-1.97)	−0.46 (−2.60-1.57)	0.09 (−1.06-1.28)	0.47
	EPBD (≤0.5), EST	−1.60 (−3.30-−0.06)	−0.72 (−2.70-1.30)	−1.25 (−2.52-−0.11)	0.44
	EPBD (>0.5-≤1), EPBD (>2-≤5)	−1.35 (−3.43-0.68)	−2.53 (−5.47-−0.11)	−1.76 (−3.41-−0.33)	0.45
	EPBD (>0.5-≤1), EST	−1.30 (−2.93-0.09)	−1.47 (−3.25-0.33)	−1.35 (−2.51-−0.35)	0.87
	EPBD (>2-≤5), EST	0.91 (−1.23-3.69)	−0.20 (−2.71-2.21)	0.43 (−1.10-1.98)	0.48
e	EPBD (>0.5-≤1), EPBD (>2-≤5)	−0.47 (−2.25-1.32)	−1.36 (−4.54-1.41)	−0.67 (−2.13-0.66)	0.54
	EPBD (>0.5-≤1), EST	0.33 (−1.42-2.03)	1.25 (−1.50-4.20)	0.48 (−0.81-1.91)	0.52
	EPBD (>2-≤5), EST	71.63 (−0.45-4.11)	0.83 (−1.66-3.26)	1.16 (−0.26-2.78)	0.6
f	EPBD (>0.5-≤1), EPBD (>2-≤5)	−0.51 (−2.01-0.95)	−0.70 (−3.41-2.06)	−0.52 (−1.80-0.71)	0.91
	EPBD (>0.5-≤1), EST	−0.14 (−1.65-1.35)	−0.01 (−2.65-2.92)	−0.10 (−1.38-1.19)	0.92
	EPBD (>2-≤5), EST	0.48 (−1.68-2.86)	0.38 (−1.74-2.46)	0.43 (−1.05-1.95)	0.94

A statistical evaluation of consistency between direct evidence and indirect evidence. There were no significant inconsistencies when P values were>0.05. a. Successful stone removal (1st session), b. Need for mechanical lithotripsy, c. Early complications, d. Post-endoscopic procedure pancreatitis, e. Late complications; f. Stone recurrence, CI: Credible interval, EPBD: Endoscopic papillary balloon dilatation, (≤0.5)/(>0.5-≤1)/ (>1-≤2)/(>2-≤5), balloon dilatation duration (min); EST: Endoscopic sphincterotomy

Publication bias

Funnel plot was drawn for PEP as an outcome indicator, and it was found that the left and right distributions of each study site were asymmetrical, suggesting the possible existence of publication bias [Figure 5].

Figure 5

Funnel plot for post endoscopic procedure pancreatitis as an outcome indicator

DISCUSSION

The efficacy of EPBD, as characteristics of included studies shows, is largely depend on size and number of stones. EPBD (balloon diameter d 10 mm) is suitable for bile duct stones with the minor diameter (<10 mm) and small number (<3), while EPLBD (balloon diameter >10 mm) can be used for stones with a larger diameter (>10 mm). Liao et al.[15] had conducted a network meta-analysis to compare short EPBD with long EPBD, demonstrating that proper extension of balloon dilatation duration could reduce complications without causing the increasing risk of PEP. However, in view of the lack of relevant literature at that time, the division of balloon dilatation duration was rough. In addition, the lack of direct comparison also affected the credibility of the analysis. Although a lot of guidelines and consensus published in recent years recommend extending dilatation duration, it was with the weak recommendation or low-level evidence.[2719] In the present study, totally 14 indirect comparisons and 4 direct comparisons were included and grouped in detail according to the balloon dilatation duration with the aim to find the optimum range of dilatation duration.

In the current study, there was no statistical difference in successful stone removal during the first endoscopic session between any two groups. Similar to previous studies, the need for mechanical lithotripsy was less common in EST compared with EPBD (>0.5, ≤1).[312] As analysis results indicated, balloon dilatation duration exerts no significant effect on stone extraction or the need for mechanical lithotripsy. Compared with EPBD, EST may have a higher success rate in the first session and less need for mechanical lithotripsy.[1226] Furthermore, EST is an effective remedial measure for EPBD.[1234] The short application time and unskilled operation of EPBD technology compared to EST may be the reasons accounting for the current phenomenon.[26] From the anatomical perspective, EST only expands the exit for stone removal, while EPBD can expand the biliary tract and papillary sphincter at the lower end of stones, making the stone removal process smoother. In the present analysis, it was seen that the stone removal efficiency of EPBD (>2, ≤5) may be slightly insufficient. In a previous study conducted by Hakuta,[18] however, the stone removal rate increased and the use of mechanical lithotripsy reduced when the balloon dilatation duration extended from 15s to 5 min. Theoretically, the extension of balloon dilatation duration contributed to Oddi sphincter relaxation and surgical site exposure, making it become easier to remove the CBD stones thoroughly. Small number of studies and inconsistent years of publication lead to low-quality evidence of this analysis. Consequently, further clinical studies are needed.

Previously, EPBD was associated with more PEP than EST, which was an important reason that limited the clinical application of EPBD.[1112] According to early ideas, PEP could be caused by the following factors: (1) pancreatic duct injury during balloon dilatation and (2) pancreatic duct obstruction and pancreatic juice regurgitation arising from papillary edema or sphincter spasm. Nevertheless, recent evidence demonstrated that the incidence of PEP could be reduced by increasing the diameter or duration of balloon dilatation.[371319] In the current analysis of PEP, EPBD (>2, ≤5) was superior to EPBD (>0.5, ≤1) and had the high probabilities of being ranked first. Although the underlying mechanism for this result does not yet remain clear, the following reasons may work.[1836] Firstly, the balloon dilatation with a small diameter balloon or short duration could result in inadequate papilla expansion, and thus the common discharge channel for bile and pancreatic juice tended to be narrow after the operation. Secondly, there was an increased chance of causing damages to pancreatic duct and its surrounding tissue, which could contribute to obstruction of the drainage of pancreatic juice. Finally, stone fragments are more likely to remain around the ampulla and lead to biliary tract obstruction. Therefore, prolonged duration of balloon dilatation to fully relax the sphincter could prevent the occurrence of PEP.

Moreover, we analysed early complications that included PEP, hemorrhage, perforation, jaundice, bile leakage and mortality. The obtained results suggested that EPBD (>2, ≤5) appeared to be superior to any other groups. In addition, we also found that it could reduce PEP without increasing the occurrence of other early complications by extending balloon dilatation duration. In the analysis of total early complications, differences between most EPBD groups (exceptf for EPBD (>2, ≤5 )) and EST group were no longer significant. This change could be attributed to the increased risk of bleeding and perforation associated with sphincterotomy. Because the advantages of EPBD in reducing bleeding and perforation have been previously reported many times,[2636] coupled with the few cases of bleeding and perforation in this analysis, separate analysis of these two outcomes was not performed. In the articles describing patients with coagulopathy, cirrhosis, altered anatomy and large periampullary diverticula, there was no statistical difference between the experimental group and the control group, while the remaining articles did not describe it. Currently, EPBD is recommended as the first choice for patients undergoing periampullary diverticula or coagulation disorders.

Cholecystitis, cholangitis, liver abscess and recurrent calculus, common late complications after endoscopic lithotomy, could be caused by cholestasis and bacterial infection in the biliary tract.[2537] In this study, there was no significant difference amongst the various comparisons as far as the late complications and stone recurrence were concerned. EST group was inclined to have a higher incidence than EPBD groups, and EPBD (>2, ≤5) had the highest probability of being the best. EST causes significant damages to the Qddi sphincter, easily leading to postoperative sphincter dysfunction.[38] Then, the reflux of intestinal contents such as digestive juices, food residue and bacteria would increase the risk of biliary tract infection and stone recurrence. Therefore, EPBD should be considered as a preferred treatment for recurrent calculus, especially patients who have previously undergone EST.[3940] To date, the available evidence on the effect of balloon dilatation duration on late complications is scant. A previous study[16] showed that the incidence of late complications not increased when the balloon dilatation duration extended from 1 min to 5 min. In addition, the results of the current network meta-analysis also demonstrated that the risk of late complications or stone recurrence did not increase when the balloon dilatation duration extending to 2–5 min. Consequently, it can be believed that the sphincter injury caused by EPBD is not notable when the balloon dilatation duration is limited within 5 min. However, the inconsistent and short follow-up time of included trails, ranging from 12 months to 6.7 years, affected the credibility of the analysis results. Longer follow-up time is needed to investigate the effect of balloon dilatation duration on late complications.

Although this analysis comprehensively collected relevant published literature and made a detailed division of balloon dilatation duration, there still remain a lot of limitations. First, the number of studies divided into each comparison, especially the direct comparison between two EPBD with different dilatation duration, was small. In addition, there was no direct comparison between EPBD (>1, ≤2) and other EPBD, which might affect the accuracy of the obtained analysis results. Second, approximately four out of five of included studies were from Asian countries, which could be one of the sources of publication bias, thus the applicability of this analysis needed to be further explored. Finally, most trials have excluded patients with coagulopathy or cirrhosis which is the subgroup tended to have the maximum benefit of the EPBD procedure. The proportions of patients with altered anatomy or large periampullary diverticula were ambiguous in many included studies, and these patients seem to benefit from EPBD procedure. Therefore, further research based on a specific population is needed.

CONCLUSION

Both EST and EPBD have obvious effects in the treatment of bile duct stones with minor diameter (<10 mm) and small number (<3). This network meta-analysis comprehensively explained and compared the effects of EPBD with different balloon dilatation duration. The extension of balloon dilatation duration exerted no significant influence on successful stone removal in the first endoscopic session or preventing the need for mechanical lithotripsy. However, it could reduce the risk of early complications, especially for PEP. Regarding late complications, prolonged balloon dilatation duration exerted no significant effect. Through this analysis, we can conclude that prolonged balloon dilatation duration could decrease early complications without increasing the difficulty of lithotomy or risk of late complications. Therefore, 2–5 min is the recommended dilatation duration range for EPBD using balloon with ≤10 mm diameter. It remains unclear whether a further extension of balloon dilatation duration would acquire better treatment outcomes. Further research based on a specific population and with longer follow-up time is needed.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.